The transfer of beef cattle feedlot manure calibrations between near-infrared spectrophotometers using three standardization techniques

Abstract

<p>The application of Near-infrared reflectance (NIR) spectroscopy in livestock manure samples has been limited by the requirement that each instrument must be individually calibrated. One possible solution to the problem is the transfer of NIR calibrations from one instrument to another. Seventy-two beef cattle feedlot manure samples were collected and scanned through the Foss NIRSystem 6500 (master) and the Foss NIRSystem 5000 (slave) instruments. Calibration equations for analyzing 11 constituents, total solids (TS), volatile solid (VS), total nitrogen (TN), ammonia nitrogen (NH ₃-N), phosphorus (P), calcium (Ca), potassium (K), sulfur (S), sodium (Na), zinc (Zn) and copper (Cu) of beef cattle feedlot manure samples were built in each instrument by the leave-one-out cross validation using partial least squared (PLS) regression. Three standardization methods including cloning, direct standardization (DS), and piece-wise direct standardization (PDS) were used to transfer the master equations to slave spectra. The 72-sample data set was split into a 30-sample standardization set to generate standardized files and a 42-sample prediction set to test the accuracy of different standardization methods. Results of this study show that the performances of calibrations for two instruments are similar. The standard error of difference (SED) was calculated based on the values of master spectra predicted by the master equations and slave spectra (standardized or not) predicted by the master equations. The SED of the standardized slave spectra was much less than the corresponding SED of the unstandardized slave spectra. The SED of the standardized slave spectra predicted by the master equations were less than the corresponding standard error of prediction (SEP) of master calibration models. This study is a first report to demonstrate that the transfer of manure sample calibrations between instruments was successful. It promises to be a satisfactory alternative to individual instrument calibration.</p>